Repeatability and Validity of Phenotypic Trait Measurements in Birds

Phenotypic trait data play a central role in ecology and evolutionary research. The quality of trait data, and the findings of subsequent analyses, depend on the quality of measurement. However, most studies overlook measurement accuracy in their study designs. We investigated the repeatability of five frequently used linear measurements of avian traits: wing length, tarsus length, bill length, bill depth and bill width and the validity of proxies for three traits: bill surface area, structural body size and tarsus size, using species from the infra-order Meliphagides (honeyeaters, fairy wrens and their allies). Repeatability varied between traits and across species for a given trait: traits larger than 13 mm showed high repeatability compared with smaller traits. By incorporating microCT technology, we showed that the formula for the surface area of a cone, a widely used proxy of bill surface area, accurately describes bill surface area within species. Surface measurement of tarsus and wing lengths were valid proxies for underlying osteology. We recommend preliminary estimation of repeatability should be undertaken for individual traits prior to data collection, in order to design suitable protocols that improve data quality, while optimizing costs involved, particularly for traits < 13 mm.